n-pass: second pass stops after 3 frames.

[hakko504]

On two occasions the last days has an encode stopped after just a few frames on the second pass (DivX5.0.3) On both occasions the start of the log file looks like this:
##map version 7
nframes 141
timescale 30000
passes 1
seq deltaT type motion_complexity texture_complexity modulation
0 0 I 0.000000 0.156536 1.000000
1 1001 I 0.000000 0.226156 1.000000
2 1001 I 0.000000 0.315195 1.000000
3 1001 P 0.133099 0.392738 1.000000
4 1001 P 0.133631 0.504824 1.000000
5 1001 P 0.133671 0.536345 1.000000
6 1001 P 0.128462 0.590566 1.000000
8 2002 P 0.198074 0.832034 1.000000
7 -1001 B 0.041805 0.124618 1.000000
10 3003 P 0.167831 0.998306 1.000000
9 -1001 B 0.058712 0.172935 1.000000
12 3003 P 0.120620 0.945845 1.000000
11 -1001 B 0.072917 0.183828 1.000000
14 3003 P 0.057667 0.587873 1.000000
13 -1001 B 0.036350 0.130677 1.000000
16 3003 P 0.061701 0.584680 1.000000
15 -1001 B 0.028340 0.130383 1.000000
18 3003 P 0.058594 0.590135 1.000000
17 -1001 B 0.039595 0.128554 1.000000
20 3003 P 0.060596 0.674427 1.000000In the other encodes I've made they always start directly with PB alternating, but in the failed encodes this doesn't start until a few frames later (frame 7/10 in the log above). I'm using the latest version of VirtualDubMod (2003-01-31), AviSynth 2.5 (2003-01-21).

Has anyone got similar problems or logfiles? (AVS posted here in another problem (http://forum.doom9.org/showthread.php?s=&threadid=43016), not with the same file though.)

[N_F]

I've had a similar problem a while ago, though I kinda doubt that's your problem.

Mine was that I by mistake overwrote the mvhints-file in between the first and second pass when I did a short quant encoding (the settings in the DivX 5.03 GUI are a bit strange, when you uncheck the log file the mvhint-file will still be written unless you uncheck, which has to be done before you uncheck the log file).

As I said, I doubt this is your problem, but it might be worth checking so that your mv-file isn't a few 100s KB.

[hakko504]

OK, I'll take a look at it. Should mention that I did restart the first pass, and got the same logfile again so it seems repeatable.

[hakko504]

That might have been it: the MV files did not look like they were correct (way too small). I'll redo these two encodes tomorrow and see what happens then.